Tri Dao
Princeton CS · Assistant Professor
Research focus: machine learning systems · efficient training and inference · sequence models · hardware-aware algorithms
保护编号 1 的自动审核：作为 ML 系统申请者，请确保您的记录中包含 GPU 内核、FlashAttention 风格算法、Mamba/状态空间模型、长上下文 sequence models、ai 的推理与加速效率，以及开源资产。
Who this advisor fits / 什么情况下适合你
- You want ML tied to real data and deployable problems (signal: machine learning systems · efficient training and inference).
- Applicants should bring systems-level ML credibility: kernels, GPU performance, long-context sequence models, state-space models, inference/training efficiency, distributed systems, and clean open-sou
- 你想把 ML 接到真实数据与落地问题（当前信号：machine learning systems · efficient training and inference）。
- 申请人应提供系统层面的 ML 可信度：内核性能 GPU、长上下文 sequence models、状态空间模型、推理与 ai 效率、分布式系统以及干净的开源代码。
What to watch for / 什么情况下要慎重
- You want pure theory ML with little domain data or cross-disciplinary collaboration.
- 你要的是纯理论 ML、几乎不碰领域数据或跨学科合作。
Public evidence as of 2026-06-10
These are decision-support signals compiled from public evidence (faculty pages, publications, lab sites) to help you ask better questions — not a ranking, rating, or allegation about this advisor. / 以上为基于公开信息整理的择校参考，帮助你提出更好的问题，并非排名、评分或对该导师的指控。
Ask a verified 学长学姐 / 同校 .edu 认证点评
The thing applicants say only 师兄师姐 can tell you — current & former students of this lab, verified by their school .edu. Open the full dossier to read or add a verified note.